Hi all,
I'm trying to get proteins predicted from novel eukaryotic DNA sequences obtained from metagenomics.
I guess protein prediction can be problematic from eukaryotic genomes....
WHAT ARE THE MAIN PROBLEMATIC ISSUES FOR PROTEIN PREDICTON?
- Intron-exon detection?
- ORF detection?
- alternative splicing stuff?
- others?
thanks!
Are you just wondering if those DNA sequences match any known eukaryotic proteins, or whether your sequences code part of a novel protein?
For the former, try blasting the sequence: https://blast.ncbi.nlm.nih.gov/Blast.cgi And see if it matches so region within a gene.
You could try tools on ExPASY https://web.expasy.org/cgi-bin/translate/dna2aa.cgi which converts your DNA sequence to an amino acid sequence, and show you the different possible open reading frames.
It's hard to say what about alternative splicing could be an issue here (without more detail about the question), but I wouldn't worry about it for now.
actually, I was wondering how can I explain (pure THEORY) that getting predicted proteins from a prokaryotic dataset should be more problematic than from an eukaryotic dataset...
thanks!
Sorry for the slip, I mean "why protein prediction should be more problematic in eukaryotes"...;-)
There could be a lot of reasons. This recent preprint may be useful: Preprint Thousands of large-scale RNA sequencing experiments yield a ...
Especially page 4, where they describe some requirements to filter out noise from potential protein coding genes. On top of what's already been mentioned, they consider minimum levels of expression (using GTEx), ORF lengths, overlapping loci. They also look at it at a transcript level.Other possible reasons that I could think of could also have to do with the technology, cost, availability of samples (e.g. sequencing of human brains for expressed transcripts), meta-genomic noise.
- Badges
- Science method